Fetal facial standard plane recognition via very deep convolutional networks

The accurate recognition of fetal facial standard plane (FFSP) (i.e., axial, coronal and sagittal plane) from ultrasound (US) images is quite essential for routine US examination. Since the labor-intensive and subjective measurement is too time-consuming and unreliable, the development of the automatic FFSP recognition method is highly desirable. Different from the previous methods, we leverage a general framework to recognize the FFSP from US images automatically. Specifically, instead of using the previous hand-crafted visual features, we utilize the recent developed deep learning approach via very deep convolutional networks (DCNN) architecture to represent fine-grained details of US image. Also, very small (3×3) convolution filters are adopted to improve the performance. The evaluation of our FFSP dataset shows the superiority of our method over the previous studies and achieves the state-of-the-art FFSP recognition results.